Data Scientist Talent Network

$100–150/hrrate as listed

Mercor is pre-screening experienced data scientists for future AI evaluation projects with frontier research labs. There is no open project today — accepted applicants sit in a talent pool and are contacted when matching work appears.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work involves

This is a talent network listing, not an active engagement. Mercor is assembling a bench of practicing data scientists so it can staff evaluation projects quickly when AI labs commission them. When work does arrive, it typically looks like judging whether an AI system can do real data science: reading a model-produced EDA notebook, statistical modeling write-up, feature engineering pipeline, or A/B test analysis, and scoring it against a rubric.

Day-to-day on a live project you would expect to:

Write task-specific grading criteria before scoring — deciding what a correct causal identification strategy, a defensible model choice, or an honest uncertainty statement actually looks like for that task
Score AI-generated (and sometimes human) deliverables against those criteria
Write the justification, not just the number — reviewers care more about your reasoning trail than your score
Absorb calibration feedback from senior reviewers and re-grade when your judgment drifts from the standard

What the platform screens for

Mercor's screen is AI-led and resume-anchored. It checks that your claimed experience is real and specific: which company, which team, what the dataset looked like, what you decided and why. Expect follow-ups that push past the summary — if you say you ran experimentation, you will be asked about power calculations, variance reduction, interference, or what you did when a test read positive for the wrong reason. The bar leans toward people who have shipped analyses at technology, research, or quantitative firms where the work was reviewed by other rigorous people.

The second thing measured is evaluation judgment. Grading is a different skill from doing: it rewards consistency, willingness to be explicit about criteria, and the discipline to score what is on the page rather than what you would have written. Vague, generous, or inconsistent scoring is the most common reason contributors wash out.

Logistics

Fully remote and asynchronous, contractor basis, project-scoped. Observed rates for this network are posted at $100–150/hr; actual rate depends on the specific project and your assessed level, and nothing is guaranteed. Because there is no immediate opening, applying is best understood as getting into the queue — some applicants hear back within weeks, others not for months. Live projects commonly run part-time at roughly 10–20 hours per week with flexible scheduling and hard turnaround windows on batches.

What they look for

1+ years professional data science experience at a leading tech, research, or quantitative firm
Strong Python and SQL for real analytical work, not just scripting
Statistical modeling, experimentation, and causal inference depth
Exceptional written technical communication — clear justifications under scrutiny
Consistency and rubric discipline when evaluating others' work
Comfort receiving calibration feedback and revising submitted scores

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Do you have at least 1 year of professional data science experience in a full-time or contract role?
Professional data science experience · required
Have you worked at a leading technology company (top-tier FAANG), AI research lab, quantitative finance firm, or equivalent organization?
Experience at leading technology or research organization · preferred
Do you have professional experience writing and shipping production or research code in Python?
Python proficiency · required
Do you have professional experience writing SQL queries to extract and manipulate data from databases?
SQL proficiency · required

0 of 4 answered

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Can you distinguish between technically correct work and work that actually answers the right question? This role requires grading whether AI systems solve real tasks, not just whether code runs.
A strong answer: A strong answer describes a specific project where you identified a gap between what was technically done and what the business needed—perhaps a model optimizing the wrong metric, or an analysis answering the wrong question—and explains how you caught and corrected it.
Where people lose it: Candidates often describe projects where they built something correctly but never show they caught misalignment between execution and intent. If you only talk about fixing bugs or improving accuracy without mentioning whether the work addressed the actual goal, you miss the core skill this role tests.